New frontiers in acute reperfusion therapy for ischaemic stroke: from recanalisation to selection,delivery and safety

Imaging-based selection and prediction of haemorrhage

Haemorrhagic transformation is the defining complication of reperfusion, and two studies in the Special Collection improve its prediction from baseline imaging. Honig et al. ¹ conducted an observational study at two Israeli centres in 398 patients with large-vessel occlusion who presented within 4 h, and all underwent baseline CT-perfusion (CTP). They compared 180 patients treated with bridging alteplase before EVT with 218 treated by direct EVT. Bridging increased the rate of parenchymal haematoma type 2 (PH-2; 6.1% vs 2.0%) without an excess of symptomatic intracranial haemorrhage. This excess in PH-2 was confined to patients with a measurable ischaemic core (any core, odds ratio 12.67; core greater than 10 mL, odds ratio 11.12). Patients without a CTP-defined core and those treated by direct EVT showed no excess, and penumbral volume was not predictive. PH-2 carried roughly four-fold higher mortality (43.8% vs 11.7%) and worse 90-day functional outcome. The authors proposed the CTP-defined core as a pre-treatment biomarker to individualise the bridging decision, while noting the observational design of their study and that their findings apply to alteplase rather than tenecteplase. Liu et al. ² addressed the extended time window, in which the balance of benefit and harm is most uncertain. In a retrospective cohort drawn from the CHABLIS-T tenecteplase trials and the Huashan Hospital registry, they studied 313 patients with anterior large-vessel occlusion or severe stenosis treated beyond 4.5 h, of whom 54 (17.3%) developed PH. For each ASPECTS region, they computed perfusion ‘mirror indices’, dividing the lesion-side value by its contralateral counterpart. Reduced cerebral blood flow and blood volume mirror indices in the lentiform nucleus independently predicted PH (adjusted odds ratios 0.07 and 0.11) and severe PH-2, after adjustment for National Institutes of Health Stroke Scale (NIHSS), ASPECTS and number of EVT passes, with an area under the curve of approximately 0.70 for PH and 0.78 for PH-2. Cerebral blood flow and blood volume outperformed Tmax and mean transit time, and the lentiform nucleus was identified as high risk because its lenticulostriate end-arteries have limited collateral reserve. The authors concluded that regional perfusion heterogeneity, rather than global metrics alone, helps stratify haemorrhagic risk in late-window reperfusion, while noting the mixed cohort and the need for external validation. Both studies show that the perfusion imaging acquired to confirm eligibility also carries quantitative information on bleeding risk, which can identify higher-risk patients before treatment.

Peri- and post-procedural haemodynamics

A second theme concerns the patient’s haemodynamic state around the procedure. Chen et al. ³ examined the association between blood-pressure variability during EVT and both functional outcome and PH. This was a post hoc exploratory analysis of the single-centre INDIVIDUATE randomised trial, which had compared individualised with standardised intraprocedural blood-pressure management during EVT under procedural sedation in 250 patients with anterior-circulation occlusion and NIHSS-score ⩾8. Several systolic blood-pressure variability metrics over the groin-puncture-to-reperfusion window were tested against 90-day functional independence and PH. Contrary to most previous reports, higher intraprocedural variability was associated with favourable outcome, both for systolic variance (adjusted odds ratio 1.002) and for average real variability (adjusted odds ratio 1.105). The authors noted that earlier studies examined the post-procedural and subacute phases rather than the procedure itself and suggested that the occluded vessel and intraluminal device may shield downstream tissue from blood-pressure swings during the intervention. Blood-pressure variability was not significantly associated with PH, and there was no interaction with the blood-pressure management strategy. The authors emphasised the exploratory, single-centre design and modest power. Baki and colleagues studied the post-procedural phase. In a retrospective cohort of 377 patients who underwent EVT for anterior-circulation occlusion and had transcranial Doppler or Duplex sonography within 72 h, an increase in ipsilateral middle cerebral artery peak systolic velocity, defined as more than 30% above the contralateral side, was strongly associated with malignant middle cerebral artery infarction, which occurred in 49 patients (13.0%). ⁴ In multivariable analysis, the adjusted odds ratio was 53.3 (95% CI 18.7–151.5), and secondary intracranial haemorrhage and higher baseline NIHSS-score were also independently associated with the outcome. The authors proposed bedside sonography as a feasible neuromonitoring tool to identify patients who may require early decompressive surgery. They noted that the relationship is partly bidirectional, since malignant swelling can itself raise measured velocities, and that an adequate temporal bone window was unavailable in roughly one-third of screened patients. Together, these studies frame the hours around recanalisation as a period that requires active haemodynamic monitoring.

Technique in anatomically challenging occlusions

The third theme addresses occlusions that were under-represented in the pivotal EVT trials. A narrative review of tandem occlusions by Duarte et al. ⁵ synthesised their diagnosis, medical and interventional management, and prognosis, treating anterior and posterior tandem occlusions as distinct entities. In anterior tandem occlusions, emergent carotid stenting improved outcomes in selected patients without an excess of bleeding. Bridging thrombolysis appeared safe and should not be withheld in eligible patients, and tenecteplase appeared at least as effective as alteplase, whereas periprocedural antithrombotic management remained highly variable with no established standard. In posterior tandem occlusions, thrombectomy was feasible and effective, but the evidence is thinner, and vertebral stenting may benefit selected patients despite high morbidity and mortality. The review also flagged carotid pseudo-occlusion as a diagnostic pitfall. A three-way comparative analysis of aspiration, stent-retriever and combined first-line approaches for basilar artery occlusion provided comparative-effectiveness data in a territory where delayed or incomplete reperfusion carries high mortality. ⁶ Across seven comprehensive centres, the investigators analysed 517 patients with isolated basilar artery occlusion treated with mechanical thrombectomy, comparing aspiration (n = 200), stent retriever (n = 260) and a combined first-line approach (n = 57) with inverse-probability weighting. After weighting, 90-day functional outcomes were broadly similar across the three approaches, but the procedural profiles differed. Stent retrievers achieved higher recanalisation than aspiration (risk ratio 1.86), the combined technique had the lowest rate of haemorrhagic transformation (risk ratio 0.39 vs aspiration) and the highest first-pass effect (73.7%), and among patients aged 80 years or older, the stent retriever was associated with better functional outcome than aspiration (39.2% vs 18%). Notably, the higher angiographic recanalisation with stent retrievers did not translate into improved clinical outcomes, a dissociation familiar in basilar occlusion and a further reminder that recanalisation is necessary but not sufficient. The authors concluded that no single technique suits all patients and that operator choice should be guided by anatomy and mechanism, pending randomised data. Both contributions reflect a field that, having established the principle of mechanical recanalisation, is now defining how best to achieve it in difficult anatomy.

Systems of care: Delivering reperfusion to the right patient, fast

The largest group of articles concerns the systems that connect patients to definitive treatment. A narrative review of the drip-and-ship model by Palaiodimou et al. ⁷ mapped current stroke-system architecture and the evidence comparing the mothership and drip-and-ship paradigms. It identified door-in-door-out time as the central modifiable metric and supports bridging thrombolysis, increasingly with tenecteplase, within transfer pathways. Two original studies add direct evidence to this framework. Gan et al. ⁸ compared EVT outcomes at thrombectomy-capable and comprehensive stroke centres and developed a web-based model to predict outcomes. Using the national ETERNITY registry of 37 Chinese centres (11 comprehensive, 26 thrombectomy-capable), they analysed patients with anterior-circulation large-vessel occlusion treated with thrombectomy within 24 h, in a derivation cohort of 975 and a validation cohort of 484. After inverse-probability weighting and adjustment, treatment at a comprehensive centre was associated with higher 90-day survival (adjusted odds ratio approximately 1.7) but no significant difference in favourable functional outcome, successful reperfusion or symptomatic intracranial haemorrhage. The authors then built an eight-variable model, deployed as a web-based tool, which showed good discrimination on external validation (C-statistic approximately 0.77 for both functional outcome and survival). They noted that the survival advantage may reflect neurocritical care and integrated services rather than the thrombectomy itself, and that the analysis was observational with validation confined to China. Stevens et al. ⁹ evaluated a smartphone application in a tertiary stroke centre. Documentation completeness rose from 15.3% to 45.6% after implementation, but treatment times did not improve, and door-to-needle and door-to-groin times trended slightly longer. At a centre with a baseline door-to-needle time below 30 min, the application had little room to shorten an already-optimised process, and its communication features were rarely used without structured training. Their findings supported the view that digital tools require integration into workflow and team training to be effective.

When thrombectomy fails: Rescue strategies

The final theme concerns the 10%–30% of thrombectomy procedures that fail to achieve reperfusion, most often because of underlying intracranial atherosclerotic disease. Anastasiou et al. ¹⁰ reported the largest analysis to date of platelet inhibition after rescue stenting, using 589 patients from 34 centres in the BASEL ICAD registry. Functional independence was achieved in 40.7% of patients, and stent occlusion occurred in 20.5%, two-thirds of these within the first 24 h. Post-procedural dual antiplatelet therapy was associated with better functional outcome, lower mortality and fewer stent occlusions than single antiplatelet therapy, whereas periprocedural glycoprotein IIb/IIIa inhibitors showed no clear association with outcome. The authors noted that the comparison is susceptible to confounding by indication, as patients with larger infarcts or periprocedural haemorrhage may have been preferentially treated with single therapy. The heterogeneity of practice they document supports the need for randomised data, which trials such as ICARUS are now beginning to provide.

Taken together, the Special Collection describes a field whose focus has moved beyond recanalisation itself. The pivotal trials established when to recanalise. The work collected here concerns what surrounds recanalisation: selection of the patients who will benefit, delivery of treatment within a narrow time window and management of the haemorrhage, oedema, reperfusion injury and reocclusion that can follow even a technically successful procedure. The same principle recurs across the imaging-prediction studies, the haemodynamic analyses, the transfer-related haemorrhage data in the drip-and-ship review and the stent-occlusion findings after rescue stenting. Several of the strongest signals, including the sonographic predictor of malignant infarction and the antiplatelet comparison after rescue stenting, are derived from observational data, and their associations should be interpreted accordingly. Their value lies in generating hypotheses and defining endpoints for future randomised and prospective studies.

The Special Collection also indicates where the next priorities lie. Distal and medium-vessel occlusion and the management of patients pretreated with anticoagulants remain comparatively underserved. Further priorities include the integration of artificial intelligence and telemedicine into prehospital triage, the definition of door-in-door-out benchmarks, prospective validation of the imaging and sonographic risk markers introduced here and randomised evidence generated within drip-and-ship and rescue-stenting pathways rather than extrapolated to them. The central message of the Special Collection is that further gains in acute reperfusion therapy will come less from the act of opening the vessel, which can now be achieved reliably, than from improved selection, faster and more equitable delivery and the management of reperfusion’s consequences.